The application of a novel fluidised photo reactor under UV-Visible and natural solar irradiation in the photocatalytic generation of hydrogen

(cid:1) Photocatalytic reduction of water using Pt-C 3 N 4 and NaTaO 3 (cid:3) La photocatalysts. (cid:1) Novel ﬂuidised photoreactor for photocatalytic hydrogen evolution from water. (cid:1) Reactor parameters eliminated mass transfer limitations to increase photocatalytic activity. (cid:1) Water reduction achieved under both UV and direct solar irradiation. (cid:1) First use of solar telescope for irradiation of photocatalytic unit. particularly important as many reactor conﬁgurations are mass transport limited, which in turn limits the efﬁciency of more effective photocatalysts in larger scale applica- tions. This paper describes the performance of a novel ﬂuidised photo reactor for the production of H 2 over two catalysts under UV–Visible light and natural solar illumination. Catalysts Pt-C 3 N 4 and NaTaO 3 (cid:3) La were dispersed in the reactor and the rate of H 2 was determined by GC-TCD analysis of the gas headspace. The unit was an annular reactor constructed from stainless steel 316 and quartz glass with a propeller located in the base to control ﬂuidisation of powder catalysts. Reactor properties such as pro- peller rotational speed were found to enhance the photo activity of the system through the elimination of mass transport limitations and increasing light penetration. The optimum conditions for H 2 evolution were found to be a propeller rotational speed of 1035 rpm and 144 W of UV–Visible irradiation, which produced a rate of 89 l mol h (cid:4) 1 g (cid:4) 1 over Pt-C 3 N 4 . Solar irradiation was provided by the George Ellery Hale Solar Telescope, located at the California Institute of Technology.

With advancements in the development of visible light responsive catalysts for H 2 production frequently being reported, photocatalytic water splitting has become an attractive method as a potential 'solar fuel generator'. The development of novel photo reactors which can enhance the potential of such catalyst, however, is rarely reported. This is particularly important as many reactor configurations are mass transport limited, which in turn limits the efficiency of more effective photocatalysts in larger scale applications. This paper describes the performance of a novel fluidised photo reactor for the production of H 2 over two catalysts under UV-Visible light and natural solar illumination. Catalysts Pt-C 3 N 4 and NaTaO 3 ÁLa were dispersed in the reactor and the rate of H 2 was determined by GC-TCD analysis of the gas headspace. The unit was an annular reactor constructed from stainless steel 316 and quartz glass with a propeller located in the base to control fluidisation of powder catalysts. Reactor properties such as propeller rotational speed were found to enhance the photo activity of the system through the elimination of mass transport limitations and increasing light penetration. The optimum conditions for H 2 evolution were found to be a propeller rotational speed of 1035 rpm and 144 W of UV-Visible irradiation, which produced a rate of 89 lmol h À1 g À1 over Pt-C 3 N 4 . Solar irradiation was provided by the George Ellery Hale Solar Telescope, located at the California Institute of Technology.

Introduction
Hydrocarbon based fuels (fossil fuels) are the primary resource for the worlds energy supplies. With increasing demand for global energy, however, there has been rapid growth in the field of energy development with a view towards the generation of an alternative energy source to provide global sustainability. Extensive focus is now being given to alternative sources such as wind, biomass, tidal, nuclear, solar, CO 2 recycling and H 2 production.
The photocatalytic reduction of CO 2 to products such as CH 3 OH and CH 4 [1][2][3][4] is a viable approach for CO 2 recycling, however is yet to be implemented on a large scale. An alternative to this is the photocatalytic splitting of water for the generation of H 2 . This is an attractive proposal as it utilises two abundant resources; H 2 O and solar illumination. The splitting of pure H 2 O into H 2 and O 2 , shown in Eq. (1), is described as an 'uphill reaction' where a large positive change in the Gibbs free energy is observed (DG o = À238 kJ mol À1 ) [5]. To achieve H 2 O splitting the conductance band of the catalyst must be more negative than the reduction potential of H + to H 2 (0 V vs NHE at pH 0), Eq. (2). The valence band must also be more positive than the oxidation potential of H 2 O to O 2 (1.23 V vs NHE at pH 0) [5], Eq. (3).
Research into photocatalytic water splitting technology has become abundant with emphasis on both catalyst and system design [5][6][7]. Systematically the concept of clean energy driven large scale photocatalytic H 2 O splitting is reliant on two factors; the efficient delivery of natural photons and the transfer of target species to photocatalyst surfaces.
Despite the simplistic concept of photocatalytic driven H 2 production, there are limited publications that investigate reactor engineering for increased H 2 production. A number of more recent publications have focused on pure water splitting in which a twocompartment or 'H-shaped' reactor is used in conjunction with a membrane to yield separate production streams of H 2 and O 2 [8][9][10][11][12][13][14]. This approach generally adopts a Z-scheme process which utilises two photocatalysts to overcome the stringent requirements faced by a single catalyst [9,14,15].
The single catalyst approach typically utilises a onecompartment reactor and will focus on increasing hydrogen production through maximising mass transfer and light distribution [16]. In view of this there are a number of advantages in developing a fluidised photocatalytic system including; improved mass transfer, excellent irradiation distribution and a high photocatalysts surface area-to-volume ratio [17][18][19][20][21]. Fluidisation is achieved as a result of the gravitational force of the catalyst particles becoming equal to the aerodynamic drag. Traditional methods of fluidisation utilise an upward stream of gas or fluid [17][18][19]21], however, previous examples have reported the use of external fields such as sound-and magnetic-assisted based fluidisation methods [20,22,23]. Examples of fluidised photocatalytic reactors have been reported using a range of model compounds including toluene [17], trichloroethylene [19], formaldehyde [20], NOx [18] and oxalic acid [21]. There are a limited number of fluidised based systems presented in the literature for water splitting or water reduction, however, in 2012 Reilly and colleagues [16] reported the construction of a UV-irradiated fluidised bed for H 2 production. The design concept of the reactor employed a low power internal illumination source (50 W) in a recirculating annular style fluidised system, which produced an optimum H 2 rate of 198 lmol h À1 over Pt loaded TiO 2 .
At the critical point of fluidisation, the light penetration of a system can be increased as the dense particle bed becomes suspended and shows fluidic characteristics. In an attempt to utilise light efficiently, previous studies have used internal illumination sources [16,17] 'light guiding' particles [17] and optical fibres [19] that can direct light deeper into the catalyst bed. Kuo and colleagues found that the addition of SiO 2 particles to their system increased light scattering and thus, toluene removal. In this investigation, a propeller-stirred fluidised system is adopted which creates cavitation of the liquid reaction medium and thus maximises light penetration.
Irradiation characteristics must be carefully considered in any photocatalytic system, particularly in a fluidised system to ensure a more efficient method for future scale up. Moreover, if photocatalytic production of H 2 is to be utilised as an alternative fuel the development of photocatalytic reactors must also focus towards systems which are capable of utilising solar radiation. A common approach used in a number of the systems reported, however, involves high power lamps, which are positioned above a reactor to provide illumination through a quartz window typically with an area of $15 cm 2 [24][25][26]. These systems utilise medium and high power lamps with a range of 300-450 W [24,[27][28][29][30]. These lamps have been used in conjunction with novel catalysts to produce a range of yields of H 2 production. In 2012, Liu and Syu [24] reported a production rate of 14.9 lmol g À1 h À1 of H 2 under irradiation from a 450 W xenon lamp. Increased rates of H 2 have been shown by Khan and Qureshi [27] and Jeong et al. [28]. Khan [31][32][33][34][35]. The system presented here utilises low power artificial irradiation sources that are easily scalable in an array structure. An alternative approach to CPC designs was adopted by utilising a solar telescope to concentrate natural illumination. This paper reports the development of a novel fluidised photo reactor suitable for use with UV-Visible and natural solar illumination. The performance of the reactor has been evaluated through monitoring the evolution of H 2 over two photo catalysts, Pt-C 3 N 4 and NaTaO 3 ÁLa. H 2 production using sustainable sources is a desirable approach in achieving clean driven fuel production. The application of low powered (36 W) visible lamps were examined with a view towards incorporation into large scale array structures. The feasibility of solar driven photocatalysis for H 2 production was demonstrated using the George Ellery Hale Telescope. This investigation aimed to demonstrate that photo reactor properties can significantly enhance the activity of a photo catalytic system, which can aid the applicability of photocatalytic technology for industrial applications.

Materials
Catalysts graphitic carbon nitride (denoted as g-C 3 N 4 hereafter), platinised carbon nitride (denoted as Pt-C 3 N 4 ) and lanthanum doped NaTaO 3 (denoted as NaTaO 3 ÁLa hereafter) were all synthesised by collaboration partners for this study. The full details of the catalyst development along with the methodology can be found in the following publications [36][37][38][39].

Catalyst characterisation
Catalyst g-C 3 N 4 was synthesized via solid state synthesis by thermal condensation of melamine at 550°C in air. Loading of the platinum co-catalyst (Pt-C 3 N 4 ) was performed by impregnation and subsequent calcination at 180°C. g-C 3 N 4 was suspended in an aqueous solution of hexachloroplatinic acid hexahydrate (1 g/L) and the solvent was evaporated at 80°C. The resulting powder was then calcined at 180°C for 2 h. NaTaO 3 :La was prepared by a conventional solid-state reaction. The starting materials, La 2 O 3 (Strem Chemicals; 99.9%), Na 2 CO 3 (J.T. Baker), and Ta 2 O 5 (Strem Chemicals; 99.8%), were mixed in the ratio Na:La:Ta = 1 À X:X:1 supposing that the sodium ions were replaced by lanthanum ions. The excess amount of sodium (5 mol%) was added in the mixture to compensate the volatilization. The mixture was calcined in air at 1170 K for 1 h and 1420 K for 10 h with intermediate grinding. The excess sodium was washed out with water after calcination. After filtration, the powder was dried at 320 K in air. The crystal structures of synthesized NaTaO:La sample was confirmed by powder X-ray diffraction (XRD) using a Philips diffractometer (X'pert Pro) with Cu-Ka radiation. Diffuse reflectance spectrum was obtained with a UV-Vis spectrometer (Shimadzu SoildSpec-3700) and was converted from reflection to absorption spectra by the Kubelka-Munk method. X-ray Photoelectron Spectroscopy (XPS) was used to observe the changes in the oxidation states of Ta and O with doping of La with Al-Ka radiation.

Reactor development
The Propeller Fluidised Photo Reactor (denoted as PFPR hereafter) was capable of providing a controllable air tight environment in which photocatalysts could be agitated to facilitate desired photocatalytic transformations in gas, solid, liquid and vapour phases. The PFPR was an annular reactor constructed from 316 stainless steel (SS) and quartz glass and consisted of a quartz glass body with SS flange and caps with a motor embedded into the base, Fig. 1. The SS flange was designed to have the quartz embedded in engineering grade, high temp resistant silicone sealant. Into this flange piece a Viton O-ring (Barnwell) groove was machined, the combination of these produced a water and gas tight seal.
The reactor environment was controlled through two Swagelok needle bonnet valves and a custom built needle extraction valve, (Fig. 1). The custom port was constructed from a Swagelok SS 316 nipple fitting consisting of a Viton O-ring (Barnwell) and GC grade septa (Hewett-Packard). This port was utilised for gas headspace needle extraction sampling. The reactor seals were tested for leaks with a Thermo Scientific GC Leak Detector. The silicone sealant also provided a failsafe had the reactor vessel become pressurised, as it will rupture before the quartz.
The unit had a calculated capacity of $210 mL with an operating capacity of 100 mL allowing a 110 mL gas headspace volume. The minimum and maximum operating capacities were 60 mL and 180 mL respectively giving a 150 mL and 30 mL gas headspace volume.
Fluidisation in the PFPR was achieved through the rotation of a 38 mm 4 blade SS 316 propeller located at the base of the reactor, Fig. 2. The propeller was driven by a 12 dcV motor, typically operating in the range 0-8 dcV by a bench top power supply. This optimum range allowed the propeller to rotate at speeds of 0-1730 rpm. This range of rotational speed gave a high level of control over both the mixing capabilities of the propeller and upward thrust required to bring stationary particles to a fluidised state in gas, vapour and liquid phases.

Artificial illumination sources
The primary source of illumination for the PFPR was an illumination box consisting of commercially available low power 36 W lamps which emitted light in the UV and visible spectrum, Fig. 3. The mirrored Perspex box was designed to house four 36 W compact non-integrated fluorescent lamps along with nine air flow control fans. The PFPR was positioned centrally in the illumination box unit to allow for full irradiation exposure.
The spectral output of the lamps is shown in Fig. 4. The lamps provided an UV emission peak at 365 nm, where the remaining strong emissions in the visible region were 435, 436, 489, 546, 588 and 611 nm.

Solar illumination
Solar illumination was provided by The George Ellery Hale Solar Telescope, positioned on the roof of the Linde and Robinson building at California Institute of Technology (CalTech), Pasadena. The telescope, which is capable of continuously monitoring and tracking the sun, was operated during the period of 24th July to the 14th August 2012. Light was delivered through a series of angled mirrors and illumination shafts. The power density, temperature and humidity of the telescope were monitored with a pyranometer. During the period of operation the mean power density was in the range of 800-900 W m 2 , the temperature ranged from 22.9 to 23.6°C and the humidity from 978 to 981 hPa. Fig. 5(a) shows the position of the telescope on the Linde and Robinson building while Fig. 5(b), shows the mirrors contained within the structure which guide light down an illumination shaft to lower levels in the building. Fig. 5(c) shows a series of 3 mirrors which are located part way down the primary illumination shaft and guide the stream of solar light through a secondary illumination shaft and out through a series of quartz windows. Fig. 5(d) shows light exciting the second illumination shaft and irradiating two PFPR units.

Photocatalytic procedure
H 2 production was performed in the PFPR over catalysts Pt-C 3 N 4 and NaTaO 3 ÁLa. In a typical experiment, a 1 g L À1 loading weight of catalyst was dispersed in 100 mL of aqueous medium containing 0.025 M oxalic acid and Milli-Q water. The PFPR was purged with Ar gas for 30 min to eliminate the presence of air in the system, during which time the rotation of the propeller was 289 rpm. After purging, the rotation of the propeller was increased to the desired level and illumination commenced. The unit was exposed to 6 h of illumination with gas head space samples taken at designated intervals.
Sample analysis was performed by gas chromatography at the University of St. Andrews and CalTech. The system used at the University of St. Andrews was an Agilent 3000 Micro GC with a dual gas carrier system (Ar and He) equipped with a TCD and molecular sieve column (HP 5890 Series II). The system used at the California Institute of Technology was a HP GC equipped with a TCD and molecular sieve (HP 5890 Series II) with N 2 as a carrier gas. Sample analysis at the University of St. Andrews was achieved by directly drawing a sample from the outlet of the PFPR via the GC internal pump. Analysis at CalTech was conducted by extraction from the injection port of the PFPR by SGE needles and directly injected into the CalTech HP GC.
The Formal Quantum Efficiency (FQE) of H 2 production in the PFPR was calculated as the reaction rate divided by the total light intensity of the UV-Visible lamps (Eq. (4)) [40]. The total light intensity irradiating the reactor surface was determined to be 396 lE cm À2 h À1 using a Lutron LX101 light meter (Lutron Electronic Enterprise Co. Ltd).

Catalyst characterisation
The catalyst g-C 3 N 4 is an organic polymer and is characterised by a layered graphitic-like structure. Its polyconjugated nature  makes it able to conduct electrons; however the orbital structure of carbon and nitrogen atoms makes it a semiconductor more than a conductive material, as is the case for graphite. This property, however, allows it to be a suitable candidate for photocatalytic processes, in particular water splitting. The layered structure of g-C 3 N 4 is reflected in the XRD pattern (Fig. 6) where the peak at 2h = 27.5°corresponds to an interlayer d-spacing of 0.323 nm. This is similar to graphite interlayer where the d-spacing is 0.326 nm. The small peak at 2h = 13.0°corresponds to an intra-layered-spacing of 0.681 nm.
For a photocatalyst to be able to perform water splitting, ideally the compound must have a band gap narrow enough for visible light absorption (E g < 3.0 eV) but also wide enough to carry out the reaction (E g > 1.23 eV). Graphitic carbon nitride fulfils these requirements. The UV-Vis spectrum (Fig. 7) of the sample shows an absorbance borderline between UV and visible irradiation (<500 nm) which translates into a band gap of 2.7 eV corresponding to the value reported in literature [21,22]. Fig. 8(a) shows a SEM image of the rough surface of g-C 3 N 4 which results in a surface area of 13 m 2 g À1 . Particle size was estimated at being, on average, 33 lm; however, bigger agglomerates were present. Layers of g-C 3 N 4 are evident in the TEM image in Fig. 8(b).
The X-ray diffraction pattern (XRD) of the synthesized NaTaO 3 doped with La (2%) was confirmed to be perovskite-type crystalline structure as evident from the 2h peaks that corresponded to the standard peak of NaTaO 3 (Fig. 9). The UV-Vis absorption spectrum of the Lanthanum doped NaTaO 3 was shown in Fig. 10(a), and the band gap of NaTaO 3 ÁLa was presented in Tauc plot calculated from its diffuse reflectance spectrum in Fig. 10(b). The absorption spectrum showed to be extended with Lanthanum doping as observed by the shoulder absorption around 340 nm. It indicated that it created a new energy level by a doped La at the upper level of the VB of non-doped NaTaO 3 . The optical band gap was calculated as 3.0 eV ( Fig. 10(b)), while the non-doped material was around 3.98 eV. Doping with La was found to be effective for both the utilisation of the light along with the charge separation of photogenerated electrons and holes compare to non-doped material.

H 2 production
The photocatalytic production of H 2 over Pt-C 3 N 4 , g-C 3 N 4 and NaTaO 3 ÁLa was monitored in the PFPR in order to evaluate the performance of the reactor. Reactor properties such as enhanced mass transport and illumination penetration and distribution were investigated in relation to photocatalytic activity. The results, shown in Tables 1 and 2, are displayed as 'total H 2 production' and 'photocatalytic H 2 production', which was required due to unexpected H 2 production observed in the control experiments.
The control experiments performed in this study showed that non-photocatalytic H 2 was evolving in the absence of a catalyst. This was concluded to be a result of the reaction between the sacrificial electron donor, oxalic acid, and SS propeller. The result of the corrosion was the formation of Fe 2+ ions on the propeller surface and subsequently the generation of H 2 (Eqs. (5) and (6)).
The propeller was constructed from SS 316, which has an austenite structure with increased levels of Cr and Mo, which should prevent corrosion. It was observed, however, that there was a significant evolution of gas under control conditions; 58.1 lmol. It has been suggested that the extent to which corrosion is promoted is dependent upon the content of carbon in the steel and concentration of Fe 2+ [41].
Typically under atmospheric conditions the corrosion of SS does not occur. Under more aggressive conditions such as acidic environments (pH < 1) the corrosion of SS has been reported [42][43][44]. While corrosion resistance under acidic conditions has been shown elsewhere in the literature [45], the results in this investigation The results obtained from the control experiments showed that the level of H 2 produced was proportional to the level of corrosion. As the level of corrosion increased, the level of H 2 increased until a point at which the formation of Fe 2+ products generated a temporary protective barrier which prevented further H 2 production. This protective barrier was eventually broken down as a result of prolonged contact with oxalic acid and corrosion continued generating H 2 [46]. Therefore, it was essential to investigate methods to prevent the continued corrosion of the propeller.
A number of conditions were investigated to minimise uncontrolled H 2 production including applying a thin polytetrafluoroethylene (PTFE) lining to the propeller surface to act as a barrier (Table 1). Table 1 shows the level of H 2 produced in the presence of Pt-C 3 N 4 as a point of comparison for the control conditions. The results showed with no PTFE lining present H 2 evolution reached 58.1 lmol after 2 h illumination which is significantly higher than the 33.1 lmol H 2 produced in the presence of Pt-C 3 N 4 . The lower level of H 2 produced when Pt-C 3 N 4 was present is likely due to the 'turbidity effect' which can occur when a catalyst is dispersed in a liquid medium. The agitation of the catalyst restricted the    Table 1 that no H 2 evolution was recorded in the absence of oxalic acid and a catalyst, however, H 2 was produced when there was oxalic present with and without illumination; 1.5 lmol H 2 under dark conditions. The level of H 2 produced in the absence of a catalyst, with and without light also showed the reaction was accelerated by the presence of photons. This aspect of the study highlights the need to conduct meticulous control experiments to ensure accurate results are obtained. In view of this fact, the production of H 2 from the corrosion of SS was monitored simultaneously in a second PFPR unit to ensure the photocatalytic activity could be accurately measured. The remaining results shown in this study as photocatalytic production were normalised to account for H 2 from corrosion.
The production of H 2 over the catalysts g-C 3 N 4 , Pt-C 3 N 4 and NaTaO 3 ÁLa was monitored and is shown in Table 2. Positive photocatalytic H 2 values are shown for all catalysts indicating that there was H 2 evolution beyond that of the control values. Pt-C 3 N 4 was shown to produce the largest volume of photocatalytic H 2 at 53.4 lmol with an evolution rate of 89 lmol h À1 g À1 . Over g-C 3 N 4 and NaTaO 3 ÁLa the evolution of H 2 was significantly lower at 5  An additional factor to be considered in the comparison of the photocatalysts is the absorption band edges of the catalysts in relation to the spectral output of the UV-Visible solar lamps (Fig. 4). While both catalysts were active under visible light, Pt-C 3 N 4 (2.7 eV) showed stronger absorption than NaTaO 3 ÁLa (3.0 eV) in the range of 400-600 nm with NaTaO 3 ÁLa showing minimal absorbance beyond 450 nm.
H 2 production over C 3 N 4 was 73.7% and 16.0% lower than Pt-C 3 N 4 and NaTaO 3 ÁLa respectively. The enhanced activity of Pt-C 3 N 4 was a result of Pt restricting the recombination of electrons and holes and catalysing proton reduction which has been well documented in the literature [26,38,47,48]. Pt has improved electron acceptor properties as a result of a larger work function (5.65 eV) as described by Mizukoshi et al. [48] which prevents the recombination of photo generated electrons and holes.
In any SED based H 2 producing system, it is also important to consider the role and pathway of the compound donating electrons. Oxalic acid was selected as the SED based on previous experiments conducted at the University of St. Andrews. As with any photocatalytic system that deploys a SED, the rate of production is limited by the depletion of the SED. Oxalic acid itself will undergo photocatalytic oxidation, ultimately resulting in complete mineralisation to CO 2 and H 2 O. The depletion of oxalic acid was evident in the results obtained as the production profiles show the rate of H 2 production decreasing with prolonged irradiation. The photodecomposition via oxidation by holes in the valence band favours molecules that are easily absorbed onto the surface (Eq. (7)).
Alternatively, however, decomposition can occur via interaction with OH Å on the catalyst surface and in bulk (Eq. (8)).
As the reactor was a sealed unit, which was purged prior to operation, the role of O 2 ÅÀ in oxalic acid decomposition was thought to be minimal. The overall mineralisation of oxalic acid is shown in Eq. (9). The final products are CO 2 and H 2 O, with few intermediates being reported in the literature.

Effect of mass transfer
While the importance of mass transfer is well understood in photocatalytic processes along with the elimination of such effects, there are few examples in the literature which investigate the impact of varying the level of mass transfer to achieve optimisation. Furthermore, there are currently no examples of a propeller based fluidised system deployed as photocatalytic reactors for hydrogen generation from water. It has been found that increasing the speed at which the propeller rotated had a number of effects; reduced mass transfer, increased light penetration and conversion of laminar flow to turbulent. The primary physical response of increasing the propeller speed was the cavitation of the aqueous medium over the range of 0-1730 rpm (Fig. 11).
The effect of propeller speed on H 2 production over Pt-C 3 N 4 using rotational speeds of 0-1730 rpm is shown in Fig. 12. The results show at a range of 0-1035 rpm, the evolution of H 2 was directly proportional to the speed of the propeller. The total production of H 2 steadily increased from 15.9 to 42.4, 55.2 and 65.7 lmol at 0, 289, 692 and 1035 rpm respectively. Thereafter, however, an increase in propeller speed to 1384 rpm showed no further increase in H 2 , with a concentration of 65.2 lmol at 1384 rpm. Moreover, once the propeller reached the maximum speed of 1730 rpm the production of H 2 reduced to 61.9 lmol. In the PFPR, the cavitation of the reaction medium began at 289 rpm which was indicated by the significant increase of 62.5% H 2 production from 0 to 289 rpm. It was also seen that as cavitation increased towards its maximum point at 1035 rpm, the rate of production increased; 23.73% and 15.98% H 2 production increase from 289 to 692 rpm and 692 to 1035 rpm respectively. As the cavitation generated at 1035 rpm was not exceeded at greater speeds, the evolution of H 2 was seen to begin to decrease: a 0.76% reduction of H 2 from 1035 to 1384 rpm and 5.06% from 1384 to 1730 rpm.
The effect of propeller speed on H 2 production can be considered with a view towards mass transfer and light penetration. At 0 rpm, the system was evidentially mass transport limited, with the reaction rate dictated by the settling velocity of the catalyst, which ultimately formed a dense bed at the base of the reactor. The significant increase in production at 289 rpm (62.5%) demonstrated the impact of reducing mass transfer limitations and increasing light penetration. At the initial point of cavitation, the dense catalyst bed is brought to a point of fluidisation as the propeller creates uniformed movement of both reactant and catalyst, generating multiple interactions on the surface of the catalyst. This relationship is further enhanced by increasing the speed of the propeller to 1035 rpm, as indicated by the H 2 production rate. Within this range, the propeller forced the aqueous reaction medium towards the wall of the unit creating full dispersion of the catalyst bed and reducing the depth of penetration required for photons to reach the catalyst surface. This feature also created maximum exposure of the aqueous medium to the 360°irradiation array.
An interesting aspect of this work was as the speed increased to 1384 and 1730 rpm, the results indicated that mass transfer limitations were no longer predominant. The plateau and further drop (5.06%) in H 2 production at 1384 and 1730 rpm highlight that the propeller speed was no longer directly proportional to production. At 1035 rpm, the aqueous medium had already reached its maximum height and thus increasing the speed of the propeller further would have no impact. Therefore, the depth of penetration required for photon adsorption at 1035 rpm was equal to that at 1384 and 1730 rpm. The level of interaction at 1384 and 1730 rpm, however, was possibly altered as a result of increased speed. The agglomeration of Pt-C 3 N 4 to form 'catalyst clusters', along with the trapping of the catalyst around the propeller may have prevented the material from being fully fluidised. As a result, mass transport limitations are increased as the available catalyst surface for photocatalytic reactions to take place was reduced. The overall result was the observed drop in H 2 production rate at increased propeller speeds. Fig. 13 shows the evolution rate of H 2 with and without Pt-C 3 N 4 over a range of rotational speeds. The highest rate of total H 2 production was 109 lmol h À1 g À1 at 1035 rpm. As it is not appropriate to display H 2 production from corrosion as per gram of catalyst, all rates shown in this figure are displayed as rate per h of irradiation. As is shown in Fig. 13 the propeller rotational speed also impacted the release of non-photocatalytic H 2 from the propeller. At increased speeds (1035 and 1384 rpm) and maximum cavitation the corrosion of SS and thus the production of H 2 was observed to decrease and plateau, 2.0 and 1.9 lmol h À1 H 2 at 1384 and 1730 rpm respectively. This was a result of a reduced contact time between the blade surface of the propeller and oxalic acid. At lower speeds (6692 rpm), however, full cavitation of the liquid medium had not occurred, which increased the residual time between oxalic acid and SS, subjecting the propeller to corrosion and thus H 2 production to occur, Fig. 13 also indicates the optimum rotational speed of the propeller to be 1035 rpm based upon levels of H 2 produced under the experimental conditions. At 1035 rpm the greatest difference in H 2 production when Pt-C 3 N 4 was present and absent was recorded at 8.94 lmol h À1 . The remaining propeller speeds were in the order 1384 rpm (8.86 lmol h À1 ) > 1035 rpm (8.4 lmol h À1 ) > 692 rpm (6.93 lmol h À1 ) > 289 rpm (5.14 lmol h À1 ) > 0 rpm (2.48 lmol h À1 ).
In order to determine the overall efficiency of the photocatalytic system, Table 3 shows the formal quantum efficiency (FQE) of H 2 produced in the PFPR over the range of 0-1730 rpm. As indicated by the reaction rate, the highest FQE of 2.25% was achieved when the propeller rotational speed was set to 1035 rpm. This calculation was a more appropriate means of displaying photocatalytic efficiency in this study as it takes into account the total light intensity of a broad spectrum source, as oppose to the photonic  efficiency calculation which is based on a monochromatic wavelength. This has previously been reported and discussed by Eggins et al. and Mills and LeHunte [40,49].
Comparison of the overall performance of the PFPR to units previously reported in the literature is difficult. Generally, the comparison of different photocatalytic systems is challenging as there are a number of varying parameters which must be taken into consideration including; catalyst properties, intensity of the illumination source, reactor geometry, pH and method of mass transfer [40]. The comparison of quantum yield and photonic efficiencies provides an indication of the overall efficiency and in occasional circumstances a platform to evaluate different photocatalytic systems. In relation to the work presented here, however, comparison of the novel PFPR to other reported reactors is challenging as the calculated FQE is not directly comparable to photonic efficiencies and apparent quantum yields reported when using a monochromatic light source. Moreover, there are currently few examples of reactors deployed for H 2 O reduction to which a comparison could be made. It is, however possible to draw some comparisons with the study conducted by Reilly and colleagues in 2012. Similarities exist between their unit and the PFPR as both adopted an annular geometric design with low power illumination, along with a simple design concept to ensure the unit was not overly complex. Reilly et al. adopted a centrally positioned internal illumination source along with a more traditional style of fluidisation through the recirculation of the liquid media, which resulted in agitation of the catalyst bed. As no photonic efficiencies were reported in this paper, cross-comparison must be based on reaction yields and rates. The Reilly et al. fluidised system produced an optimum H 2 rate of 198 lmol h À1 which was significantly higher than the hourly rate of H 2 produced in the PFPR ($8.9 lmol h À1 ). To accurately compare these systems however, the following reaction conditions must be considered; catalyst loading, illumination time frame and source. The fluidised system described by Reilly et al. used a high catalyst loading weight of 80 g, along with UV light and 24 h irradiation. Interestingly, if the rate of H 2 production was to be displayed as lmol g À1 h À1 , as it often is, the rate would be $2.475 lmol g À1 h À1 , which was significantly lower than that of 89 lmol g À1 h À1 produced in the PFPR. Furthermore, the catalysts deployed in PFPR were novel synthesised for excitation under UV-Vis and natural illumination, while the Pt-loaded TiO 2 catalyst developed by Reilly et al. was restricted to UV light.

Preliminary solar H 2 production over NaTaO 3 ÁLa
If photocatalytic production of H 2 is to be utilised as an alternative fuel it is essential that the development of photocatalytic reactors focuses towards systems which are capable of converting solar energy. Solar driven photocatalysis has been reported in the literature primarily for the applications of water detoxification [50,51], CO 2 reduction [52,53] and H 2 production [31,32,35]. To date, however, there are no examples of solar H 2 production from the use of a solar telescope to concentrate the stream of photons. Jing et al. [7] discussed a number of parameters which restrict the application of solar driven photocatalysis including the inability to efficiently utilise natural light. One such limitation of solar driven photocatalysis is the requirement to first concentrate the stream of photons, which can result in a change in temperature and subsequently a change in pressure. In using the solar telescope these limitations are eliminated which prevents a thermal reaction taking place over that of a photocatalytic one. Solar H 2 production over NaTaO 3 ÁLa (which was the available catalyst at the time of testing) in the PFPR is shown in Table 4.
Under entirely natural solar light a total concentration of 4.6 lmol H 2 was generated which gave a total production rate of 8 lmol h À1 g À1 . The evolution of H 2 from the corrosion of the propeller was also monitored simultaneously in a second PFPR unit and was found to be lower at 1.7 lmol, 63.0% lower than when NaTaO 3 ÁLa was present. Therefore, a calculated photocatalytic H 2 evolution rate of 5 lmol h À1 g À1 was achieved under concentrated solar illumination.
The total production of H 2 over NaTaO 3 ÁLa and the photocatalytic production of H 2 under UV-Visible and natural solar light are also summarised in Table 3. While H 2 production is lower under solar light, the results remain promising. In order for photocatalysis to progress there is a need for both solar activated materials and reactors that can maximise their efficiency. The PFPR quartz body provided a large surface area for solar irradiation whilst the propeller increased mass transfer to increase the activity of the catalyst.
While the production of H 2 under solar illumination was lower than under artificial light, there are factors which must be considered; the electromagnetic spectrum of natural and artificial light, the photon delivery system of the solar irradiation source and variable nature of solar irradiation itself. The solar telescope used in this preliminary investigation continually monitored and tracked sun light during the day and concentrated it through a series of gold plated mirrors and illumination shafts. This allowed for a concentrated stream of solar photons to be guided to the tubular body of the PFPR. It has been reported by Jing et al. 2009 [32] that the concentration of solar light is essential, stating that 'a solar reactor cannot be designed without first concentrating natural sunlight in an attempt to enhance the intensity'. As the electromagnetic spectrum of solar light has a broad spectrum in comparison to the narrow range emitted from high power lamps, which are typically adopted in photocatalytic systems, the intensification of the illumination source is essential. The concentration of natural light in this circumstance was achieved while maintaining atmospheric conditions and room temperature which ensured the production of hydrogen was photocatalytic and not thermally activated.
While the mirrors and shafts of the solar telescope concentrated the irradiation stream, the light path was subject to contaminants. Bickley et al. [54] stated that the number of particles and contaminants between the source of exposure and the catalyst should be kept to a minimum. Under UV-Visible light, the source of illumination is typically immersed or externally positioned in close Table 3 The calculated FQE for the production of H 2 in the PFPR over the propeller rotational speed range of 0-1730 rpm. The FQE was determined using Eq. (4) and a total light intensity of 396 lE cm À2 h À1 . proximity to the reactor and as such prevents the presence of contaminants in the light path. In comparison the light path of the George Ellery Hale telescope is approximately 40 ft, and is exposed to various environmental contaminants. The weather dependency and variable nature of solar light as an irradiation source is also a key factor. Cloud coverage and unexpected weather can reduce and restrict the delivery of light through the telescope. The result of which can be dark periods which reduces the photocatalytic efficiency.

Conclusions
In summary, the production of H 2 over Pt-C 3 N 4 and NaTaO 3 ÁLa under UV-Visible light and natural solar irradiation was achieved in a novel fluidised photo reactor. The reactor properties were also shown to enhance the photocatalytic activity of the system with an increase from 24 lmol h À1 g À1 at 0 rpm to 89 lmol h À1 g À1 at 1035 rpm. At increased speeds of 1384 and 1730 the rate of H 2 produced decrease indicating mass transport limitations no longer predominant at speeds greater than 1035 rpm. A limitation to the current design of the PFPR is the uncontrolled production of H 2 as a result of SS corrosion by oxalic acid. The corrosion was localised to the propeller and was reduced by operating the propeller at 1035 rpm and applying a PTFE coating to the surface. The evolution of H 2 over NaTaO 3 ÁLa under solar illumination was also achieved with a production rate of 5 lmol h À1 g À1 . These initial results demonstrate the potential for solar water reduction for H 2 production from a sustainable source. This is a key area of focus for current ongoing work being conducted.